Update app.py

app.py

@@ -6,9 +6,10 @@ import time
 import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
 import streamlit as st
-
+import pytube as pt
 from transformers import AutoModelForSeq2SeqLM
-import torch
+MODEL_NAME = "drinktoomuchsax/whisper-small-hi"
+lang = "en"
 
 from threading import Thread
 os.environ["COQUI_TOS_AGREED"] = "1"
@@ -20,6 +21,33 @@ HF_TOKEN = os.environ.get("HF_TOKEN", None)
 # Set the device to GPU or CPU
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
+
+### Whisper Start
+
+pipe = pipeline(
+    task="automatic-speech-recognition",
+    model=MODEL_NAME,
+    chunk_length_s=30,
+    device=device,
+)
+pipe.model.config.forced_decoder_ids = pipe.tokenizer.get_decoder_prompt_ids(language=lang, task="transcribe")
+
+# Transcription function
+def transcribe(file):
+    text = pipe(file)["text"]
+    return text
+
+# YouTube transcription function
+def yt_transcribe(yt_url):
+    yt = pt.YouTube(yt_url)
+    stream = yt.streams.filter(only_audio=True)[0]
+    stream.download(filename="audio.mp3")
+    text = pipe("audio.mp3")["text"]
+    return text
+
+### Whisper END
+
+
 # Load the tokenizer and model
 tokenizer = AutoTokenizer.from_pretrained("merve/chatgpt-prompts-bart-long")
 model = AutoModelForSeq2SeqLM.from_pretrained("merve/chatgpt-prompts-bart-long", from_tf=True).to("cuda" if torch.cuda.is_available() else "cpu")
@@ -130,15 +158,24 @@ with right:
     st.markdown('''<h3><i class="fa fa-pencil"></i> Form 2</h3>''', unsafe_allow_html=True)
 
     # Box 3: Form 2
-    prompt2 = st.text_input("Enter Prompt", key="prompt2")
-    image_url2 = st.text_input("Enter Image URL", key="image_url2")
-    if st.button("Submit Form 2", key="submit2"):
-        payload = {"prompt": prompt2, "image_url": image_url2}
-        response = requests.post("https://d3ndnam3-hf.space/api", json=payload)
-        if response.status_code == 200:
-            st.write(f"**Response:** {response.json().get('response', 'No response')}")
+    uploaded_file = st.file_uploader("Upload an audio file", type=["mp3", "wav", "flac", "aac"])
+    microphone_input = st.audio("Record audio using microphone", format="audio/wav")
+    if st.button("Transcribe"):
+        if microphone_input and uploaded_file:
+            st.warning("WARNING: You've uploaded an audio file and used the microphone. The recorded file from the microphone will be used, and the uploaded audio will be discarded.")
+            file_to_transcribe = microphone_input
+        elif microphone_input:
+            file_to_transcribe = microphone_input
+        elif uploaded_file:
+            file_to_transcribe = uploaded_file
         else:
-            st.write("Failed to get a response")
+            st.error("ERROR: You have to either use the microphone or upload an audio file")
+            file_to_transcribe = None
+
+        if file_to_transcribe:
+            with st.spinner("Transcribing..."):
+                transcription = transcribe(file_to_transcribe)
+            st.write(transcription)
 
     # End of Box 3 and third Carousel Item
     st.markdown('''<h3><i class="fa fa-pencil"></i> Form 3</h3>''', unsafe_allow_html=True)